Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The treatment of drug abuse focuses on prevention of the development of addiction, elimination of existing addiction and suppression of symptoms associated with drug withdrawal (Jaffe, 1987). Current pharmacotherapies usually target specific neurotransmitter systems, which are presumed to mediate the effects of a given class of abused substances. These "targeted," specific pharmacotherapies are used in spite of striking similarities concerning the long-term consequences produced by all of known drugs of abuse such as drug addiction. Moreover, opioid antagonists and agonists are used in the treatment of opioid abuse despite the fact that some therapies (e.g., naltrexone) are not only ineffective, but, in addition, are consistently refused by addicted individuals. In opioid substitution therapy (e.g., with methadone), Ball and Ross (1991) reported a recidivism rate after discontinuation of treatment of ~80%. Dopaminergic pathways, particularly of the mesolimbic system, have been associated with the motivational (reinforcing) aspects of drug abuse. However, to date, antiaddictive therapies targeting dopaminergic neurotransmission have failed to produce a therapeutic breakthrough (Pulvirenti and Koob, 1994).

The pharmacological terms tolerance, sensitization, dependence and withdrawal characterize the clinical term "drug addiction." Drug tolerance refers to the state in which repeated administration of the same dose of a given drug elicits a diminishing effect or the need for an increasing drug dose to produce the same effect. Sensitization (reversed tolerance) refers to the opposite situation in which repeated administration of the same drug dose elicits an escalating effect. Chronic treatment with many drugs, including opioid agonists, produces gradually developing dependence that is defined as the need for continued reexposure to avoid a withdrawal syndrome. The withdrawal syndrome is characterized by physical and motivational disturbances that are the expression of ongoing drug dependence. The term "the maintenance of opioid dependence" refers to the situation in which one can precipitate withdrawal syndrome a long time after the administration of opioid agonist has been discontinued. Contemporary theories of drug addiction consider addictive substances as potent reinforcers. In turn, chronic exposure to these reinforcing drugs can lead to drug addiction, which is best defined as compulsive drug seeking and taking behavior despite adverse consequences. Because such effects are common for all of drugs of abuse (despite the diversity of their chemical structures and plethora of acute pharmacological actions), it is believed that the common neurobiological substrates mediate the motivational impact of drugs producing drug addiction (Eddy et al., 1965; Goudie, 1991; Robinson and Berridge, 1993).

The NMDA receptor complex is a ligand-gated cationic channel, which consists of a pentameric assembly of subunits that contains several regulatory sites. Thus, apart from a recognition site for a primary transmitter (glutamate) it includes sites sensitive to glycine (which is obligatory for receptor activation), polyamines, zinc, redox state and protons (Danysz et al., 1995a; Kutsuwada et al., 1992). It should be noted that agents acting at these sites have not yet been deeply investigated in clinical settings. NMDA receptor function can also be inhibited by blockade of the open ion channel (Danysz et al., 1995a; Kutsuwada et al., 1992). In contrast to the other NMDA antagonists, several open channel (uncompetitive) blockers have been tested in clinical settings, and some of them are being used for medical purposes. Unfortunately, the majority of the high-affinity open channel NMDA receptor antagonists, e.g., phencyclidine or MK-801 (dizocilpine), produce side effects, including confusion, psychotomimetic activity and memory disturbances (Luby et al., 1959; Troupin et al., 1986), that render them unacceptable for a clinical use. In contrast, low-affinity, strongly voltage-dependent uncompetitive NMDA receptor antagonists have been suggested to produce less side effects, perhaps because of weaker effects when the level of NMDA receptor activation is in the physiological range (Chen et al., 1992; Parsons et al., 1993; Rogawski, 1993). One such low-affinity NMDA channel blocker, memantine, has been used clinically for many years in Europe, and is apparently devoid of these side effects if dosed properly, i.e., in gradually increasing doses until therapeutic concentrations are achieved (Ditzler, 1991; Görtelmeyer and Erbler, 1992). A similar favorable therapeutic profile holds true for other agents of this kind, such as amantadine and dextromethorphan (Danysz et al., 1995a; Rogawski, 1993). It seems that the use of NMDA antagonists possessing a relatively credible clinical profile (memantine, dextromethorphan) in novel applications (in this case, the treatment of drug abuse) should be preferred over the use of novel agents that have no such long clinical history.

Converging lines of evidence indicate the essential involvement of NMDA receptors in phenomena related to drug addiction. In preclinical studies, NMDA receptor antagonists decrease tolerance to the locomotor effects of alcohol (Khanna et al., 1993) and sedatives (File and Fernandes, 1994), attenuate sensitization (reverse tolerance) to stimulants (Pudiak and Bozarth, 1993; Wolf and Khansa, 1991) and modify adaptive changes caused by nicotine treatment (Shoaib and Stolerman, 1992). Moreover, NMDA receptor antagonists affect opioid tolerance and dependence processes. Several studies indicated that NMDA receptor antagonists decrease tolerance to the analgesic effects of opiates (Ben-Eliyahu et al., 1992; Bhargava and Matwyshyn, 1993; Elliott et al., 1994; Kolesnikov et al., 1993, 1994; Tiseo and Inturrisi, 1993; Tiseo et al., 1994; Trujillo and Akil, 1991). Similarly, reduction of the physical as well as motivational aspects of the expression of morphine dependence (measured by naloxone-precipitated morphine withdrawal syndrome) has been shown by many investigators (Cappendijk et al., 1993; Higgins et al., 1992; Popik et al., 1995; Rasmussen et al., 1991; Tanganelli et al., 1991; Trujillo and Akil, 1991). In addition, NMDA receptor antagonists inhibit the development of morphine dependence (Elliott et al., 1994; Tiseo and Inturrisi, 1993; Tiseo et al., 1994; Trujillo and Akil, 1991) as well as its maintenance (Popik and Skolnick, 1996). "Antiaddictive" effects of NMDA receptor antagonists (dextromethorphan, ibogaine) have been reported also in initial clinical trials (Koyuncuoglu and Saydam, 1990; Lotsof, 1995).

In spite of intensive efforts, the mechanism of the inhibitory effects of NMDA receptor antagonists in various measures modeling drug addiction remains highly speculative. This is in part caused by the fact that the molecular and physiological mechanism(s) underlying drug dependence and addiction are per se not well understood. Although drug dependence (defined by withdrawal syndrome) is traditionally considered to be a major factor in the maintenance of compulsive drug use (Eddy et al., 1965), contemporary theories stress the importance of the reinforcing (incentive) properties of abused substances (Robinson and Berridge, 1993) in the development and maintenance of drug addiction. Thus, it might be hypothesized that antiaddictive treatments should inhibit ongoing drug-seeking through a decrease of the reinforcing impact of drugs of abuse, rather than by attenuating solely the severity of the withdrawal syndrome.

Recently, Bespalov et al. (1994) demonstrated that the nonselective glutamate receptor antagonist, kynurenic acid, attenuates the acquisition and expression of CPP induced by morphine. The same treatment inhibits morphine-induced facilitation of responding in the electrical intracranial self-stimulation paradigm (Bespalov et al., 1994). Similar attenuation of the development of morphine-induced CPP has been recently shown for MK-801 (dizocilpine) and CGP-37849, which are uncompetitive and competitive NMDA receptor antagonists, respectively (Tzschentke and Schmidt, 1995). The ability to induce CPP and to facilitate intracranial self-stimulation, as well as to induce and maintain self-administration, defines in animal models reinforcing properties of drugs and, according to several theories of drug addiction, potential abuse in humans (for review, see Goudie, 1991). Thus, it seems likely that treatments decreasing these reinforcing actions of drugs in animal models would be effective in diminishing intake of drugs of abuse in humans.

CPP offers a reliable measure for assessing the reinforcing value of pharmacological treatments and other reinforcers, including food. In the case of pharmacological manipulations, during training a drug supposed to have reinforcing properties is paired with a distinctive compartment, whereas its vehicle is paired with the other compartment of the same training apparatus. After completing the association phase, drug-free animals explore both compartments. It is assumed that an increase of the time spent in a compartment associated with a given treatment reflects the incentive (positively reinforcing) value of that treatment (Bindra, 1978), whereas a decrease suggests aversive properties (Carr et al., 1989; Mucha et al., 1982). Numerous studies indicate that in addition to the reinforcement, associative learning must occur for the development of CPP (e.g., White and Carr, 1985). CPP induced by opiates is stereospecific, can be blocked by opioid antagonists and is probably mediated by action at mu rather than kappa opioid receptors (Mucha and Herz, 1985). In addition, dopaminergic mesolimbic pathways have been reported to participate in opioid-induced CPP (Spyraki et al., 1983).

The objective of the present experiments was to determine whether the NMDA receptor antagonist memantine could affect the reinforcing impact of morphine as well as the motivational aspects of morphine withdrawal in a CPP paradigm. Memantine was selected for the present studies, because it is a use-dependent NMDA receptor antagonist (IC₅₀= 0.5-3.0 µM [Bormann, 1989; Chen et al., 1992; Kornhuber et al., 1989; Parsons et al., 1993] that currently is clinically used in Germany in the treatment of senile dementia and spasticity (Ditzler, 1991; Görtelmeyer and Erbler, 1992). More specifically, memantine's effects on both acquisition and expression of CPP induced by morphine were assessed in the first experiment. The specificity of memantine's effect was tested subsequently in the CPP paradigm in which rats were reinforced by food, rather than morphine injection. The next experiment was carried out to find if memantine may influence the motivational aspect of naloxone-precipitated morphine withdrawal, as measured in the CPA paradigm. Because the impact of memantine on acquisition/expression of CPP could also be explained by its effects on learning and memory, experiments were performed in which memantine was given to rats acquiring and retrieving spatial information in the water maze. In addition, because the effects of memantine on morphine-induced CPP could be interpreted as a diminution of the subjective (interoceptive) effect of morphine, in the last experiment we investigated the influence of memantine on the interoceptive cue produced by morphine in a drug-discrimination paradigm.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Subjects. Male Wistar rats (approximately 300 g of weight at the beginning of the experiment) were housed under standard laboratory conditions for at least 2 weeks before experiments started. Animals were kept in plastic cages, four rats per cage (58 × 37 × 19 cm) in the animal room with a controlled light-dark cycle (lights on, 7:00 A.M.; off, 7:00 P.M.). Water and commercial food were available ad libitum, unless otherwise stated.

Drugs. Morphine HCl (Polfa), naloxone HCl (Endo), memantine HCl (Merz and Co.), (+)-morphine and etonitazene (both gifts of Dr. K. Rice, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD) were dissolved in physiological saline. Saline was used as placebo. The doses of morphine, naloxone and etonitazene correspond to the doses calculated as base for all other agents as respective salts. All injections were given in a volume of 1 ml/kg i.p., except in the drug discrimination experiment in which all agents except memantine were injected subcutaneously.

Induction of CPP. The CPP procedure was similar to that described previously by Papp and Moryl (1994). Four identical wooden boxes with white and black chambers (30 × 20 × 25 cm each) were used. The chambers had distinct floor textures (plain wood in the white chamber and wire mesh in the black chamber, respectively). The gray central area (12 × 20 × 25 cm) constituted a "neutral" chamber. The CPP procedure consisted of adaptation, pretest, acquisition of a conditioned response and the posttest. During the first 3 days of training (adaptation phase), the rats were placed individually in the apparatus to freely explore it for 10 min daily. These trials as well as preliminary studies indicated that almost all subjects preferred the black chamber over the white one. On day 4 (pretest), the time spent in the white chamber during a 10-min free exploration session was measured and recorded. This measure was used as an initial preference score for each subject.

Induction of CPA. The CPA procedure was conceptually based on the work of Higgins and colleagues (1991, 1992). Rats were injected with morphine (10 mg/kg, twice daily, at 9:00 A.M. and 5:00 P.M.) for 8 days. On days 3, 4 and 5 of morphine treatment, adaptation sessions (10 min of unobserved, free exploration) were carried out at least 2 hr after the morning dose of morphine. On day 6 of morphine treatment, the pretest session was performed, as described for the CPP procedure, with the exception that the time in the black chamber was measured. On day 7, at least 90 min after the morning injection of morphine, rats were injected with placebo and 30 min later again injected with placebo. Immediately after the second injection, subjects were placed in the white chamber of the apparatus for 30 min. On day 8, at least 90 min after the morning injection of morphine, rats were injected with vehicle, 3.75 or 7.5 mg/kg of memantine, and 30 min later they received a 0.1 mg/kg naloxone injection. Immediately after the naloxone injection, rats were placed in the black chamber of the apparatus for 30 min. No more morphine injections were given to the animals. On day 9, the posttest was performed, during which the time spent in the black chamber was recorded during a session lasting for 10 min. The number of rats in each group was 10 to 15.

Effects of memantine on the acquisition and retention of spatial learning and memory. Rats were trained to find a metal platform that was submerged 1 cm below the water surface in the swimming pool (50 cm high, 180 cm in diameter) (Popik et al., 1994a,b). The platform was positioned half-way between the wall and the center of the circular pool and remained in this position throughout the 4 training days. There were six trials on each training day. Each trial started from one of the four compass points around the pool perimeter, with the sequence, e.g., N, E, S, W, N, E. Rats were gently placed into the water, facing the wall; the latency to find the platform was measured for each rat. Subjects were kept on the platform for 30 sec, after which the next trial started. After completion of the six swimming trials, rats were transferred to the "drying" cage and later to their home cages. The experimental room contained numerous visual cues and had dispersed illumination that allowed videotaping.

Effects of memantine on morphine-produced interoceptive cue as measured in a drug-discrimination paradigm. The Morris water maze as described above was used to train rats (n = 8) to discriminate between placebo (physiological saline) and ()-morphine HCl interoceptive cue (3.5 mg/kg). Rats were injected with placebo or morphine 20 to 30 min before being gently placed in the tank at either NE or SW compass starting point. Subjects were given four trials per day, usually 6 to 7 days a week. The sequence of starting points was random throughout the experiment and independent from treatment conditions; however, in a given day a sequence of either the NE, SW, NE, SW or SW, NE, SW, NE of starting points was used. During the training, only one submerged platform was present inside the tank and only one treatment was given. Half of the subjects were required to associate placebo injection with the platform positioned in the NW quadrant and morphine injection with the platform positioned in the SE quadrant. For the remaining half of the subjects, the assignment of platforms was reversed. Typically, a double alternation scheme of training was used, with rats receiving placebo-placebo-morphine-morphine injections on subsequent days. Subjects were kept on the platform for 30 sec, after which the next trial started. After completion of the four swimming trials, rats were transferred to the "drying" cage and later to their home cages. It took approximately 40 training days for a rat to achieve the criterion of "good" performance, defined as eight out of nine consecutive first correct swimming trials. A trial was considered "correct" if the rat swum from the starting point to the respective platform without swimming in the vicinity (an area about two times bigger than the size of platform) of the platform associated with the alternative treatment (see fig. 1 for details).

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Images of the typical swimming paths of rats trained to discriminate morphine from placebo in the Morris water maze. Presented are layouts of the experimental setup. Smaller rectangles represent actual platform positions, with shaded smaller rectangles representing "correct" and open smaller rectangles representing "incorrect" platform positions, respectively. The bigger dashed rectangles around "incorrect" platform represent area that, if encountered by a subject, indicated erroneous trial. A and B images represent error-free trials, and C and D images represent erroneous trials.

Data analysis and statistics. In the CPP and CPA studies, the preference scores were expressed as a percent increase or decrease of time spent by a rat in a given chamber of the apparatus on pretest and posttest. Data were analyzed by ANOVA, followed by Duncan or Student-Newmann-Keuls tests.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of memantine on the CPP induced by morphine. During the CPP pretest there were no differences in the time spent in the white chamber of the apparatus among groups (ANOVA, P = .23). However, the preference scores to that chamber after conditioning varied greatly among groups (F(8,97) = 4.09, P = .0003). Rats injected with morphine in the white (initially nonpreferred) chamber demonstrated a marked preference for this chamber during the drug-free posttest. To assess the effects of memantine on the acquisition of CPP, rats were injected with memantine before every conditioning session.

View larger version (48K): [in this window] [in a new window]   Fig. 2.   The effects of memantine on acquisition and expression of morphine-induced CPP. Presented are preference scores for the white chamber of the apparatus after 6 days of conditioning. Open bar indicates scores of placebo-conditioned control subjects (i.e., rats that received 6 placebo injections during conditioning and placebo injection before the posttest). The double hatched bar represents scores of morphine-conditioned control subjects. The hatched bars represent scores of rats that received memantine, at doses indicated, before every conditioning trial and placebo before the posttest (left panel), or placebo before every conditioning trial and memantine before the posttest (right panel). The dotted bar represents scores of rats conditioned with 7.5 mg/kg of memantine instead of morphine. The number in each group is 10 to 15 rats. Values are mean ± S.E.M. Symbols: #P < .05 vs. morphine-conditioned subjects; P < .05 vs. placebo-conditioned controls; *P < .05 vs. the respective group, Student-Newman-Keuls test.

Effects of memantine on the CPP induced by food. Rats that received placebo injections both during conditioning and before the posttest acquired CPP to the chamber associated with food (fig. 3).

View larger version (37K): [in this window] [in a new window]   Fig. 3.   The effects of memantine on acquisition and expression of food-induced CPP. Presented are preference scores for the white chamber of the apparatus after 6 days of conditioning. Open bar indicates scores of placebo-conditioned control subjects (i.e., rats that received 6 placebo injections during conditioning and placebo injection before the posttest). Hatched bars represent scores of rats that received memantine, at doses indicated, before every conditioning trial and placebo before the posttest (left panel), or placebo before every conditioning trial and memantine before the posttest (right panel). The number in each group is 10 to 15 rats. Values are mean ± S.E.M. ANOVA indicated no effects of memantine treatment.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   The effects of memantine on food intake during acquisition of food-induced CPP. Presented are mean ± S.E.M. number of pellets consumed by rats during conditioning days 2, 4 and 6. On these tests, rats were offered five food pellets in the white chamber of the apparatus. Rats received placebo (circles) or memantine at doses 3.75 (triangles) or 7.5 (squares) mg/kg, respectively, 30 min before being inserted into the apparatus. Symbols: *P < .05 vs. placebo-treated (Student-Newmann-Keuls test, after one-way ANOVA); #P < .05 vs. memantine 7.5 mg/kg treated on day 2 (paired t test).

Effects of memantine on the CPA induced by naloxone-precipitated morphine withdrawal. Placebo-pretreated, naloxone-challenged control rats demonstrated ~75% of their initial preference to the black chamber, whereas morphine-dependent naloxone-challenged animals showed significantly lower preference to that chamber (fig. 5). ANOVA demonstrated significant differences among groups (F(4,51) = 2.91; P = .031).

View larger version (39K): [in this window] [in a new window]   Fig. 5.   The effects of memantine on acquisition of naloxone-precipitated CPA in morphine-dependent rats. Presented are preference scores for the black chambers of the apparatus after 2 days of conditioning. The double-hatched bar indicates the scores of control rats that were not morphine-dependent and received placebo injections during conditioning. The open bar represents the scores of morphine-dependent control subjects. The hatched bars represent scores of morphine-dependent rats that received memantine at the doses indicated on the bottom of the figure. The number in each group is 10 to 15 rats. Values are mean ± S.E.M. Symbols: *P < .05 vs. control, morphine-nondependent, placebo-treated rats, Student-Newman-Keuls test.

Morris water maze procedure: Effects of memantine on acquisition and retention of spatial learning and memory. All placebo-treated rats trained in the Morris water maze demonstrated rapid acquisition of the spatial memory (fig. 6). Two-way repeated measures ANOVA performed on the swimming latencies data for all 24 trials demonstrated significant effects of the treatment (F(2,483) = 16.54; P < .001), trial number (F(23,483) = 26.88; P < .001) and interaction (F(46,483) = 1.92; P < .001). Significant differences among groups (memantine vs. vehicle) were detected when data were analyzed separately for the first (P < .001), second (P < .001) and third (P < .01), but not the fourth day of training.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Effects of memantine on the acquisition and retrieval of the spatial learning task. Presented are mean escape latencies of rats pretreated with 7.5 mg/kg of memantine (squares) or placebo (circles or triangles) 30 min before each of the first learning trials (i.e., trials 1, 7, 13 and 19). Asterisks indicate that on a given trial rats treated with memantine demonstrated significantly longer latencies to find the platform than placebo-treated subjects on the given trial (P < .05, Student-Newman-Keuls test, after one-way ANOVA). No differences were detected between the placebo-treated groups. S.E.M. is not shown for the clarity of the figure.

Effects of memantine on morphine-produced interoceptive cue as measured in the drug-discrimination paradigm. Rats tested the day after the morphine training dose responded to the graded doses of morphine with an ED₅₀ of 0.74 mg/kg, whereas on the day after placebo training the ED₅₀ was 1.23 mg/kg (see table 2 for raw data). These dose-response curves were not different from each other as calculated with the Litchfield and Wilcoxon (1949) procedure. Therefore, the ED₅₀ of the combined data from both groups was 0.98 mg/kg with confidence limits of 0.64 to 1.51 mg/kg.

View this table: [in this window] [in a new window]   TABLE 2 Morphine-produced drug discrimination in rats trained to discriminate ()-morphine from placebo in the Morris water maze

View this table: [in this window] [in a new window]   TABLE 3 Substitution to etonitazene and (+)morphine in rats trained to discriminate ()-morphine from placebo in the Morris water maze

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The present findings demonstrate that memantine in a dose-dependent fashion attenuated the acquisition and expression of morphine-induced CPP, but had no significant effect on the acquisition or expression of CPP induced by food presentation. Moreover, memantine partially inhibited acquisition of motivational aspects of naloxone-precipitated morphine withdrawal as measured in the CPA paradigm. In addition, at the dose that blocked morphine-induced CPP, memantine by itself produced neither CPP nor CPA. We failed to demonstrate that memantine at doses effective in CPP and CPA paradigms modified the interoceptive cue produced by morphine. Although, initially, memantine attenuated the acquisition of spatial learning, this effect was tolerated after 3 days, and no interference with the retrieval of spatial memory was found in the present experiments.

Consonant with previous studies (for reviews see Carr et al., 1989; Goudie, 1991) the present experiments revealed that morphine produced clear-cut CPP. Memantine inhibited both the acquisition and expression of morphine-induced CPP. Similar inhibition was found previously for kynurenic acid (Bespalov et al., 1994), a nonselective antagonist of excitatory amino acid receptors. Recently, Tzschentke and Schmidt (1995) demonstrated that dizocilpine and CGP 37849, uncompetitive and competitive NMDA receptor antagonists, respectively, attenuate the acquisition of morphine-induced CPP.

The reduction of morphine-induced CPP by an NMDA receptor antagonist (memantine) found in the present study could be related to several effects that were not explicitly ruled out in the previous studies. First, uncompetitive NMDA receptor antagonists are known to exert inhibitory effects on associative learning (see Danysz et al., 1995b, for review). Because the establishment of CPP involves associative learning, it is likely that treatments inhibiting plastic processes would inhibit CPP as well. However, it is generally agreed that NMDA receptor antagonists affect the acquisition of new information but not the storage or recall of associations that are well established (Caramanos and Shapiro, 1994; Danysz et al., 1995b; Shapiro and Caramanos, 1990). Moreover, the amnestic effects of memantine are usually seen at higher doses (10-20 mg/kg; Misztal et al., 1995), which suggests that the learning impairment is not of major importance for antagonism of morphine-induced CPP in which a lower dose was used. Although memantine initially increased latencies to find the hidden platform (which likely reflects disturbances in swimming behavior and/or initial attenuation of spatial learning acquisition), these effects were not longer apparent on the last day of training (fig. 6). Thus, even if memantine were initially to impair the acquisition of morphine-induced CPP, the purported amnestic actions of this compound, by analogy, should cease by day 3 of conditioning, whereas the association phase in the CPP experiments lasted for 6 days. Further evidence against strong amnestic effects comes from the analysis of swimming paths recorded during the probe trial that specifically measures the strength of spatial memory. Thus, table 1 demonstrates no differences between spatial navigation behavior of rats treated with memantine or placebo during the acquisition of the Morris spatial task. In agreement with our data, Barnes et al. (1996) reported that rats treated chronically with food containing memantine (30 mg/kg/day) show increased maintenance of long-term potentiation in vivo and normal learning in the Morris maze. Other studies indicate that memantine infused at doses leading to serum concentrations observed in humans has no effects in naive animals and actually improves radial maze learning in rats after entorhinal cortex lesion (Zajaczkowski et al., 1996). Moreover, Ditzler (1991), as well as Görtelmeyer and Erbler (1992), reported positive cognitive effects of memantine in demented patients.

Memantine at the dose inhibiting acquisition of morphine-induced CPP did not significantly affect food-induced CPP (fig. 3), which suggests that memantine-treated rats were able to associate food reinforcement with a distinctive environment. It should be mentioned that although statistical analysis (ANOVA, P > .05) indicated that memantine affected neither acquisition nor expression of food-induced CPP, the data presented in fig. 3 suggest that inhibitory effects of memantine might have been detected if a different statistical approach had been used. Such inhibitory effect may likely be related to the initial decrease of food intake seen in memantine-treated rats (fig. 4) and/or initial attenuation of learning processes as revealed in the Morris water maze (fig. 6). It remains an open question whether these factors contributed to memantine-induced attenuation of acquisition of morphine-induced CPP (fig. 2). Nonetheless, these factors are unlikely to explain the inhibitory effects of memantine on the expression of morphine-induced CPP, because memantine did not affect the retrieval of spatial memory (table 1).

It has been suggested recently that the treatments attenuating the rewarding aspects of drugs of abuse may have aversive properties by themselves, which confounds the interpretation of the results of CPP studies. Such aversive effects were demonstrated for the L-type calcium channel blockers (Pizzi and Cook, 1996) which inhibit, e.g., cocaine-induced CPP (Pani et al., 1991). To rule out this interpretation, the experiment was performed in which memantine (7.5 mg/kg) was administered to the rats in the black compartment of CPP apparatus. If memantine would have aversive actions, it is likely that rats would avoid the memantine-associated chamber. The data presented under "Results" demonstrate the lack of aversive effects of memantine, thus making the "aversive" interpretation unlikely.

Memantine could also decrease morphine-induced CPP by attenuating the interoceptive cue produced by morphine and perhaps related to its reinforcing action. Morphine possesses clear discriminative properties (Colpaert, 1977; Shannon and Holtzman, 1979; Young et al., 1992) that are thought to contribute to its abuse liability (Jaffe, 1987; Martin and Jasinski, 1969). The correct choice of the drug-associated compartment in the CPP procedure is based, among others, on the recognition of spatial cues, whereas the most often used procedures for measuring discriminative properties of drugs are based on operant conditioning. Spatial learning and operant conditioning tasks use different forms of memory; and, perhaps because of this fact, they are affected differently by amnestic treatments (Danysz et al., 1995b; Wessinger, 1994). It was therefore reasonable to compare the effects of memantine on morphine-induced CPP in a task relying on similar cognitive processes. Thus, for the present experiments a drug-discrimination procedure has been developed that is based on the ability of rats to associate the spatial position of a hidden platform with the interoceptive cue produced by the drug. Spatial learning paradigms based on water T-maze learning have been used previously for studying discriminative effects of drugs (Henriksson and Jarbe, 1972; Jarbe, 1987).

Several earlier studies indicate that the cue produced by morphine in drug discrimination paradigms arises from central drug actions and is independent from its peripheral effects (Colpaert et al., 1975; Gianutsos and Lal, 1975, 1976). In the drug discrimination procedure used in the present study, rats were able to discriminate 3.5 mg/kg of morphine from placebo, a finding in accordance with numerous previous studies that used operant conditioning techniques. To check the reliability of the present version of the drug-discrimination paradigm, we investigated whether the discriminative effect of morphine: 1) is similar 1 day after morphine treatment as well as 1 day after placebo treatment; 2) is stereospecific; 3) if another opioid mu agonist will substitute for the morphine interoceptive cue and 4) if the discriminative effect of morphine could be blocked by a specific antagonist. The data presented (tables 2 and 3) demonstrate that the water maze could be used successfully as a tool for studying discriminative effects of drugs. Thus, 1) the ED₅₀ value for the morphine interoceptive cue appeared not to be different the day after the morphine training dose from the day after placebo treatment; 2) (+)-morphine, an unnatural isomer that is devoid of opioid effects (Adams et al., 1991; Van der Kooy et al., 1982) did not substitute for the ()-morphine interoceptive cue; and 3) etonitazene, a mu agonist ~1000 times more potent than morphine in behavioral and neurochemical assays (Rice et al., 1983; Wikler et al., 1963) substituted for the morphine interoceptive cue with respectively higher potency than morphine itself. In addition, a low dose of naloxone abolished the discriminative cue produced by morphine. However, subjects injected with morphine and memantine behaved similarly to rats injected with morphine only (table 4). Thus, the present data suggest that it is unlikely that memantine decreased the morphine interoceptive cue and thereby inhibited acquisition of morphine-induced CPP. In agreement with our data, Bespalov and colleagues (1995) reported recently in abstract form that a nonselective glutamate receptor antagonist, kynurenic acid, did not affect the heroin discriminative cue.

Since most probably, memantine did not inhibit the acquisition of morphine-induced CPP through learning impairment or attenuation of the morphine interoceptive cue, it is likely that it attenuated the reinforcing impact of morphine. This observation is supported by the fact that memantine also inhibited the expression of morphine-induced CPP, which suggests that this compound diminished the conditioned (secondarily reinforcing) properties of the morphine-associated environment. The inhibitory action of NMDA receptor antagonists on the reinforcement produced by morphine are far from being fully understood. Nevertheless, the inhibitory effects of memantine on morphine-induced CPP but lack of effect on morphine discrimination may be explained by differential involvement of dopaminergic transmission in these two phenomena. For example, it is generally agreed that opioid (morphine and heroine)-induced CPP depends critically on dopamine mesolimbic systems, whereas no specific central site has been ascribed a primary role in the morphine discriminative cue (cf. Joharchi et al., 1993). Moreover, Spyraki et al. (1983) found that disruption of dopaminergic transmission by 6-hydroxydopamine-induced lesions of the nucleus accumbens or by haloperidol treatment blocks CPP induced by opioids including morphine. On the other hand, no agreement exists on the involvement of dopaminergic transmission in the mediation of interoceptive properties of opioids. Thus, complete generalization to a morphine cue after amphetamine pretreatment has been reported in some (Shannon and Holtzman, 1979) but not all studies (Joharchi et al., 1993). Similarly, it appears that dopaminergic antagonists do not block the opioid discriminative stimulus (Colpaert et al., 1976; Colpaert, 1977).

The differential involvement of dopaminergic pathways in opioid-induced CPP and drug discrimination of opioid compounds prompt consideration of an interaction between glutamatergic and dopaminergic neurotransmission as a potential basis of the inhibitory effects of memantine on morphine-induced CPP. Several findings indicate that the mesolimbic dopaminergic pathways are the anatomical substrate of drug reinforcement (Di Chiara and Imperato, 1988; Phillips and Le Paine, 1980; also see Koob, 1992; Robinson and Berridge, 1993 for reviews) where dopaminergic and glutamatergic transmission interact mutually (Jedema and Moghaddam, 1994; Kalivas et al., 1989; Krebs et al., 1991; Moghaddam and Bolinao, 1994; Wolf et al., 1994). In the striatum and nucleus accumbens glutamatergic afferents can increase dopamine release through interaction with NMDA receptors (Di Chiara and Imperato, 1988; Krebs et al., 1991). Increased locomotor activity has been observed after intra-accumbens injection of glutamate; this effect is blocked by dopaminergic antagonists (Donzanti and Uretsky, 1983). Because mesolimbic dopaminergic transmission is inevitably involved in the regulation of drug reward (for reviews see Koob, 1992), it is likely that its modulation may affect the reinforcing properties of opioids. It remains intriguing why the noncompetitive NMDA receptor antagonists that have stimulant actions and increase activity of the mesolimbic system, as evidenced by biochemical (Bubser et al., 1992) and electrophysiological data (French, 1994), evoke inhibitory effects on the reinforcing effects of morphine.

In contrast to another uncompetitive NMDA receptor antagonist, dizocilpine that by itself produces clear CPP (Layer et al., 1993; Papp and Moryl, 1994), memantine seemed devoid of these reinforcing effects (fig. 3). This result was unexpected, but it should be noted that dizocilpine exerts this reinforcing effect in a narrow dose range (Layer et al., 1993); and, as already mentioned, the blockade of NMDA receptor channel by memantine is characterized by much faster kinetics (Chen et al., 1992) and stronger voltage dependence (Parsons et al., 1993). In general, it can be argued that lower affinity directly translates into faster channel blocking kinetics and, within a certain range, into a more favorable side-effects profile. However, too low affinity usually results in a loss of selectivity which, in turn, can result in an increase of side effects (see Parsons et al., 1995 for discussion). The dose of memantine effective in the present study (7.5. mg/kg) would be expected to produce serum concentrations of ~1.5 µM (see Danysz et al., 1994 for discussion), and a somewhat (40-50%) lower concentration would be predicted to occur in the cerebrospinal fluid (Danysz et al., 1994). In demented patients treated with memantine, serum concentrations of ~0.4 to 0.5 µM are observed (Kornhuber and Quack, 1995). Based on patch-clamp studies (Parsons et al., 1993) such concentrations should be sufficient to inhibit NMDA receptors in the brain. Hence, the failure to demonstrate CPP after memantine was not caused by insufficient dosing, which is also supported by the finding indicating that neuroprotective activity is observed at doses leading to serum levels of ~1 µM (Misztal et al., 1996; Wenk et al., 1995).

Converging lines of evidence indicate that NMDA receptor antagonists are effective in inhibiting the physical aspects of the expression of morphine dependence (Cappendijk et al., 1993; Popik et al., 1995; Rasmussen et al., 1991; Tanganelli et al., 1991; Trujillo and Akil, 1991). Such effects have also been demonstrated recently for memantine at doses expected to affect primarily, if not solely, NMDA receptors (Popik and Skolnick, 1996). It is worth noting that the inhibitory effects of memantine on the expression of morphine withdrawal was attenuated by glycine administration (Popik and Skolnick, 1996), which suggests the involvement of NMDA receptors. The expression of morphine dependence is typically measured by precipitating morphine withdrawal syndrome with an opioid antagonist. Although contemporary theories of drug addiction do not attribute a major motivational role of physical withdrawal syndrome to the maintenance of opioid dependence (Robinson and Berridge, 1993; Wise and Bozarth, 1987), the abrupt cessation of opioid administration produces a state of extremely unpleasant sensations that may motivate addicts to maintain illicit drug use (Eddy et al., 1965). The difference in physical and psychological aspects of withdrawal syndrome is illustrated by the poor long-term effectiveness of clonidine treatment of abstinence syndrome in humans. Clonidine, an alpha-2 adrenergic agonist, decreases the opioid abstinence syndrome in addicted individuals (Gold et al., 1978); however, it principally alleviates the physical but not psychological (motivational) consequences of opioid withdrawal (Charney et al., 1981; Jasinski et al., 1985). It might be hypothesized that the poor overall outcome (i.e., success rate of approximately 40% [Rounsaville et al., 1985]) of clonidine treatment of opioid abuse is caused by the fact that clonidine fails to affect the motivational aspects of opioid withdrawal syndrome. Therefore pharmacological manipulations that would diminish the motivational aspects of opioid withdrawal may be considered as more beneficial treatments of opioid dependence. Such inhibitory effects on motivational aspects of opioid withdrawal syndrome as demonstrated in the CPA procedure have been shown previously for dizocilpine (Higgins et al., 1992) and for memantine in the present experiments (fig. 5). Clinical trials with other uncompetitive NMDA receptor antagonists, dextromethorphan and ibogaine, in opioid addicts were reported to be successful (Koyuncuoglu and Saydam, 1990; Koyuncuoglu, 1995; Sheppard, 1994).

In conclusion, the present study demonstrates that in rats memantine may attenuate the reinforcing aspects of morphine and of morphine withdrawal syndrome, being inactive by itself in the CPP or CPA paradigms. It remains to be established if the diminution of the physical (Popik and Skolnick, 1996) as well as motivational (present study) signs of expression of opioid withdrawal produced by memantine also can be seen in humans.